1. Volume 74, Issue 5, Pages 887-898 (June 2012)
Mg2+ Block of Drosophila NMDA Receptors Is Required for Long-Term Memory Formation and CREB-Dependent Gene Expression 
Tomoyuki Miyashita, Yoshiaki Oda, Junjiro Horiuchi, Jerry C.P. Yin, Takako Morimoto, Minoru Saitoe 
Neuron 
Volume 74, Issue 5, Pages (June 2012)
DOI: /j.neuron
Copyright © 2012 Elsevier Inc. Terms and Conditions

2. Figure 1
Physiological and Pharmacological Properties of Endogenous dNMDARs in the Fly Brain
(A) The pupal neuronal primary culture system. (Left) GFP image of neurons in an intact elav/GFP pupal brain. (Middle) A phase contrast image of a neuron dissociated from the pupal brain (upper) and its GFP signal (lower). (Right) Whole-cell clamping of a GFP-positive neuron. See also Figure S1 for dNR1 distribution.
(B) Inward currents induced by 100 μM NMDA in GFP-positive cells are abolished in the presence of 20 mM Mg2+. GFP-positive cells responded to NMDA at 0 mM Mg2+.
(C) Both NMDAR antagonists APV (10 μM) and MK-801 (10 μM) significantly decrease NMDA-activated currents in GFP-positive neurons. ∗p < 0.05 by t test. n = 6 for all data. Error bars in all figures in this paper indicate SEM.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

3. Figure 2
Suppression of dNMDAR Mg2+ Block in Neurons from Transgenic Pupae Overexpressing dNR1(N631Q)
(A) Schematic diagram of dNR1 and amino acid sequence comparisons of the TM2 domains of Drosophila (dNR1 and dNR2), mouse (NR1, NR2A, and NR2C), and C. elegans (NMR1 and NMR2) NMDAR subunits. The white circle (upper) and shadowed box (lower) indicate the Mg2+ block site in the TM2 domain.
(B) Current-voltage (I-V) curves generated from neurons from elav/dNR1(wt) (n = 6) and elav/dNR1(N631Q) pupae (n = 6). Currents elicited by 100 μM NMDA were normalized to peak responses at +50 mV. Due to Mg2+ block, all examined neurons from elav/dNR1(wt) displayed a typical J-shaped I-V relation in the presence of 20 mM extracelluar Mg2+.
(C) (Left) I-V relationships in high Na+ (open circle) and high Ca2+ (closed circle) extracellular solutions for elav/dNR1(wt) and elav/dNR1(N631Q) neurons. I-V relationships in each extracellular solution were recorded from the same neuron after confirming the presence [for elav/dNR1(wt)] or absence [for elav/dNR1(N631Q)] of Mg2+ block. (Right) Reversal potentials in high Na+ extracellular solution (Vrev,Na) and in high Ca2+ extracellular solution (Vrev,Ca) were measured from the left panel, and relative Ca2+ permeabilities (PCa/PNa) were calculated using the Goldman-Hodgkin-Katz (GHK) equation.
See also Figures S2 and S3.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

4. Figure 3
Transgenic elav/dNR1(N631Q) Flies Are Defective for Long-Term Memory but Not Learning
(A) In contrast to dNR1EP3511 (EP3511) and dNR1EP331 (EP331) hypomorphs, learning (LRN) is normal in elav/dNR1(wt) and elav/dNR1(N631Q) flies.
(B) Short-term memory is normal in elav/dNR1(wt) and elav/dNR1(N631Q) flies and reduced in dNR1EP3511 (EP3511) and dNR1EP331 (EP331) hypomorphs.
(C) One-day memory after spaced training is reduced in elav/dNR1(N631Q) flies and in dNR1 hypomorphs.
(D) One-day memory after massed training is normal in elav/dNR1(N631Q) flies and in dNR1 hypomorphs.
In all panels, data from wild-type flies are indicated by the solid (mean) and dotted (SEM) lines. ∗∗∗p < , as determined by one-way ANOVA and Bonferroni post hoc analyses. n = 8 to 10 for all data.
See also Figures S3 and S5.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

5. Figure 4 Mg2+ Block Is Physiologically Required for LTM Formation
(A) LRN assayed using a short-duration training paradigm. LRN PI scores increase progressively as a function of training duration. Mg2+ block mutations do not inhibit LRN. Rather, elav/dNR1(N631Q) flies have slightly improved LRN compared to other transgenic control flies at training durations of 10 and 20 s. ∗p < 0.05 and ∗∗p < 0.01, n = 8–12 for all data.
(B) Expression of a dNR1(N631Q) transgene restores normal memory in dNR1EP331 mutants. ∗∗p < 0.01, n = 6–8 for all data.
(C) Acute feeding of 1 mM RU486, starting one day prior to training, significantly impairs LTM formation in elavGS/dNR1(N631Q) flies. ∗∗p < 0.001, n = 8–12 for all data.
(D) Expression of a dNR1(N631Q) transgene in the ellipsoid body of the central complex, Feb170/dNR1(N631Q) and C42/dNR1(N631Q), and in the MBs, OK107/dNR1(N631Q) and C309/dNR1(N631Q), significantly impairs LTM formation. ∗∗p < 0.001, n = 10 for all data.
See also Figure S4.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

6. Figure 5
LTM-Dependent Induction of activin, homer, and staufen Expression
Increased transcription of activin, staufen, and homer is observed 6 hr after spaced training. Expression is significantly higher after spaced training than after massed training. Expression of RP-49 (ribosomal protein), dlg, and ζ are the same after spaced and massed training. Expression of all genes was normalized to expression of GAPDH1. Comparison of gene expression 3 and 24 hr after spaced versus massed training gave similar results (data not shown). One-way ANOVA indicates significant differences due to training protocol for activin, homer, and staufen. ∗∗p < as determined by Bonferroni post hoc analyses, n = 10 for all data.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

7. Figure 6
NMDARs and Mg2+ Block Are Required for LTM-Dependent Increases in activin, homer, and staufen Expression
(A) The increase in transcription of activin, homer, and staufen 6 hr after spaced training is suppressed in hypomorphic dNR1EP3511 (EP3511) flies. Two-way ANOVA indicates significant differences due to training, genotype, and interaction between training and genotype. ∗∗p < determined by Bonferroni post hoc comparisons. n = 10 for all data.
(B) The increase in transcription of activin, homer, and staufen after spaced training in wild-type and transgenic control elav/dNR1(wt) flies is abolished in elav/dNR1(N631Q) Mg2+ block mutant flies. Two-way ANOVA indicates significant differences due to training, genotype, and interaction between training and genotype. ∗∗p < n = 10 for all data.
(C) Spaced training causes increases in HOMER protein in various brain regions, including the calyces of the MBs (Cas), protocerebral bridge (PB), the antennal lobes (ALs), and lateral protocerbrum (LP). These increases do not occur in brains of elav/dNR1(N631Q) flies. Restricting dNR1(N631Q) expression to the MBs prevents increases in HOMER protein in the Cas (yellow arrowhead) but does not affect expression in the PB (yellow arrows).
See also Figure S6.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

8. Figure 7
Mg2+ Block Is Required for CREB-Dependent Induction of LTM-Associated Genes
(A) The increase in activin, homer, and staufen transcription after spaced training does not occur in heat-shocked hs-dCREB2-b (17-2) flies. Two-way ANOVA indicates significant differences due to genotype, training, and interaction between genotype and training. ∗∗p < 0.01 and ∗p < 0.03, determined by Bonferroni post hoc comparisons, n = 8–12 for all data.
(B) Basal expression of dCREB2-b transcripts is significantly increased in elav/dNR1(N631Q) flies. One-way ANOVA indicates significant differences due to genotype. ∗∗∗p < compared to +/+ and elav/dNR1(wt) flies, n = 10 for all data. The wild-type level of dCREB2-b expression (normalized to GAPDH1 expression) was defined as 1.
(C) dCREB2-b protein is significantly increased in elav/dNR1(N631Q) flies but not in dNR1EP3511 flies. dCREB2-b protein was normalized using tubulin, and the wild-type amount was defined as 1. One-way ANOVA indicates significant differences due to genotype. ∗∗∗p < compared to +/+ and elav/dNR1(wt) flies, n = 10 for all data.
(D) Comparison of dCREB2-b/dCREBtotal ratios. The ratio of repressor to total dCREB2 (dCREB2-b/dCREBtotal) is increased more than 2-fold in elav/dNR1(N631Q) flies compared to wild-type and elav/dNR1(wt) flies. One-way ANOVA indicates significant differences due to genotype. ∗∗p < 0.01, n = 10 for all data.
(E) Removal of external Mg2+ increases dCREB2-b protein in cultured brains. A greater than 2-fold increase in dCREB2-b protein was observed in single brains dissected from adult flies, cultured in the absence of external Mg2+. This increase was suppressed by the NMDAR antagonist MK801 and by the dNR1EP3511 mutation. TTX was added to all cultures to suppress action potentials. One-way ANOVA indicates significant differences due to genotype. ∗∗p < 0.01 and ∗p < 0.05, n = 6 for all data.
See also Figure S7.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions

9. Figure 8
Mg2+ Block of dNMDARs Functions to Suppress dCREB2-b Expression
(A) Correlation between amounts of dCREB2-b protein and defects in LTM. Wild-type (+/+) and hs-dCREB2-b flies were heat shocked at 35°C for various durations (min) and separated into two groups. One group was used for immunoblotting to quantify the amounts of dCREB2-b protein (upper panel), and the second group was subjected to spaced-training. (Lower panel) LTM scores (one-day memory after spaced training) are plotted against the amounts of dCREB2-b protein in flies heat shocked for the indicated times. The dCREB2-b/tubulin ratio was normalized to that of wild-type without heat shock. As seen from the regression line obtained from wild-type and hs-dCREB2-b flies, the amount of dCREB2-b in elav/dNR1(N631Q) flies is sufficient to disrupt LTM formation.
(B) A model for Mg2+ block function in LTM formation. Ca2+ influx during correlated activity activates adenylyl cyclase and kinases, including CaMKs, ERK, and PKA, promoting associative learning and dCREB2-dependent induction of LTM-associated genes. Ca2+ influx during uncorrelated activity, driven by unpaired stimuli and by “minis,” is usually prevented by Mg2+ block. In the absence of Mg2+ block, low Ca2+ influx from unpaired stimuli and minis increases a repressor isoform dCREB2-b, preventing formation of LTM.
See also Figure S7.
Neuron  , DOI: ( /j.neuron )
Copyright © 2012 Elsevier Inc. Terms and Conditions